Detail View

Markerless 6-DoF Pose Estimation for Closed-Loop Control of Tendon-Driven Continuum Manipulators

Citations

WEB OF SCIENCE

Citations

SCOPUS

Metadata Downloads

Title
Markerless 6-DoF Pose Estimation for Closed-Loop Control of Tendon-Driven Continuum Manipulators
Alternative Title
텐던 구동 연속체 매니퓰레이터의 폐루프 제어를 위한 마커리스 6 자유도 자세 추정
DGIST Authors
Chunggil AnMinho Hwang
Advisor
황민호
Issued Date
2026
Awarded Date
2026-08-01
Type
Thesis
Description
Markerless 6D Pose Estimation, Visual Servoing, Sim-to-Real, Differential Rendering, Self-supervised learning
Table Of Contents
List of Contents
Abstract i
List of contents ii
List of tables iv
List of figures v

Ⅰ. Introduction 1
Ⅱ. Related Works
2.1 Pose Estimation of Continuum Manipulators 4
2.2 Synthetic Data Generation for Continuum Manipulators 4
2.3 Differentiable Rendering-Based Pose Estimation 5
2.4 Closed-Loop Control of Continuum Manipulators 5
Ⅲ. Hardware and Kinematics
3.1 Forward Kinematics 6
3.2 Jacobian and Inverse Kinematics 8
Ⅳ. Synthetic Data Generation
4.1 URDF Modeling of the Continuum Manipulator 9
4.2 Simulation Environment and Rendering 9
4.3 Simulation-Driven Ground-Truth Generation 10
4.4 Domain Randomization 12
4.5 Dataset Composition 12
Ⅴ. Pose Estimation Framework
5.1 Region-of-Interest Extraction 14
5.2 Multi-Feature Fusion Network 15
5.3 Keypoint Detection Module 15
5.4 Segmentation Module 16
5.5 Pose Regression Head 16
5.6 Stereo-Aware Attention Module 17
5.7 Feed-Forward Rendering-Based Refinement Module 18
5.8 Loss Formulation and Multi-Stage Training 19
Ⅵ. Model Component Ablations
6.1 Comparative Performance on Synthetic Data 22
6.2 Effect of the Refinement Module 22
6.3 Effect of the Stereo Configuration 23
6.4 Time Efficiency and Model Complexity 24
Ⅶ. Sim-to-Real Transfer via Calibration and Self-Supervised Adaptation
7.1 Real Camera Intrinsic and Extrinsic Calibration 25
7.2 Self-Supervised Real Adaptation with Pseudo Ground Truth 27
7.3 Effect of Jaw Case Attachment on Pose Estimation Accuracy 31
Ⅷ. Model Validation in Real Environments
8.1 Performance with and without Self-Supervised Adaptation 33
8.2 Performance Comparison with Previous State-of-the-Art Methods 34
Ⅸ. Closed-Loop Control of the Continuum Manipulator
9.1 Jacobian-Based Position-Based Visual Servoing 36
9.2 Point-Reaching Tasks 38
9.3 Trajectory Tracking Tasks 38
Ⅹ. Conclusion 40
Reference 42
URI
https://scholar.dgist.ac.kr/handle/20.500.11750/60785
http://dgist.dcollection.net/common/orgView/200001007046
DOI
10.22677/THESIS.200001007046
Degree
Master
Department
Department of Robotics and Mechatronics Engineering
Publisher
DGIST
Show Full Item Record

File Downloads

  • There are no files associated with this item.

공유

qrcode
공유하기

Total Views & Downloads

???jsp.display-item.statistics.view???: , ???jsp.display-item.statistics.download???: